To the Editor: The effects of highly active antiretroviral therapy (HAART) on thyroid function are unclear at present. Although it is noted that the majority of patients with HIV disease display normal thyroid function,1,2 there is an increasing awareness of a growing number of patients receiving HAART presenting with symptoms of thyroid disease requiring treatment.3 Conflicting results concerning the necessity of routine thyroid screening in HIV patients receiving HAART have been observed.4-6 A recent study from the Royal Free Hospital, London, United Kingdom, suggested that screening is not warranted due to an observed low frequency of thyroid abnormalities in their cohort,4 whereas the longitudinal study performed by the University of Sassari, Italy, indicates that screening was recommended.3 The Chelsea and Westminster Hospital HIV cohort in London is the largest single center HIV cohort in Europe, and information is prospectively collected on the individuals who attend. All patients in this study attended between April 1995 and June 2006 and had at least 1 blood test for thyroid function in the absence of a defined clinical protocol. Hyperthyroidism was defined as thyroid-stimulating hormone (TSH) level below normal (TSH normal range 0.3-4.4 mU/L) and free thyroxine (T4) or free triiodothyronine above that of the normal range (free T4 normal range 9.0-26.0 mU/L). Hypothyroidism was defined as TSH level above normal range and free thyroxine (T4) or free triiodothyronine below that of the normal range (normal range 2.23-5.3 pM). To establish incidence, person-days of follow-up were converted to person-years at risk. To keep the coefficient of the person-years at risk constant, this was log transformed and used as the offset in the Poisson regression before analysis using the Genmod procedure in SAS version 9.1 with loge link and Poisson error distributions. A total of 2437 individuals had routine thyroid function tests performed. The incidence of hyperthyroidism was 3.4 (95% confidence interval: 1.5 to 6.8) per 10,000 patient-years and of hypothyroidism it measured 10.7 (95% confidence interval: 6.9 to 15.8) per 10,000 patient-years. Of these 2437 individuals, 54 (2.2%) were identified with abnormal thyroid function, of whom, 26 were diagnosed with hyperthyroidism and 28 with hypothyroidism. Twenty-one (80.8%) of 26 hyperthyroid patients were male and 5 (19.2%) were female. Twenty-two (78.6%) of 28 hypothyroid patients were male and 6 (21.4%) were female. The clinical prevalence of hyperthyroidism in our HIV cohort was calculated to be 1.01% and that of hypothyroidism was 1.2%. Of the patients tested, thyroid antibodies were present in 40% of those with hypothyroidism and 66.7% of patients with hyperthyroidism, a finding that may suggest an association with immune restoration after HAART. Table 1 represents treatments that 54 patients demonstrating abnormal thyroid function were receiving in comparison with the entire clinical cohort. Of the patients receiving nonnucleoside reverse transcriptase inhibitors, 13 (81%) of the hypothyroid patients and 7 (81%) of the hyperthyroid patients were receiving efavirenz, which was statistically significant when compared with the total population (P = 0.025). Twelve of the patients who developed hyperthyroidism were prescribed protease inhibitors, which was also significant when compared with the entire population (P = 0.002). No significant associations were observed with simple confounding and residual factors of age, gender, race, nadir CD4 cell count, and time on antiretroviral therapy with thyroid function.TABLE 1: Treatments Patients With Thyroid Disease Were Receiving Compared With the Entire Clinical CohortWe have revealed a higher than expected incidence of hypothyroidism in a large cohort of HIV-infected individuals although the retrospective nature of our cohort with an absence of a defined clinical protocol for thyroid function testing results in selection bias. From these data, we found that hypothyroidism was most commonly associated with protease inhibitors and hyperthyroidism with nonnucleoside reverse transcriptase inhibitors, efavirenz in particular. As patients with HIV infection are now typically treated with HAART for decades, the metabolic complications of treatment require further exploration. Based on the previous extensive work in this area, and these data, we recommend routine screening of thyroid function in HIV-infected patients receiving HAART. The association of thyroid disease including subclinical thyroid test abnormalities with different treatments requires further investigation. Mark Nelson Tom Powles Abigail Zeitlin Priya Sen Andrew Scourfield Mark Bower Brian Gazzard Justin Stebbing The Department of HIV Medicine The Chelsea and Westminster Hospital Imperial College School of Medicine London, United Kingdom Funding source: none declared. Conflict of interest: none. All authors had full access to the data.